Blender VSE Pipeline

Overview

Automate video editing with Blender's Video Sequence Editor (VSE) using Python. Add and arrange strips (video, image, audio), create transitions, apply effects, assemble edits from file lists, and render final video — all headlessly from the terminal. Checked against the Blender 5.2 LTS Python API reference.

Blender 5.0 renamed the VSE API: Sequence became Strip, scene.sequence_editor.sequences became .strips, sequences_all became .strips_all, and new_effect takes length, input1, input2 instead of frame_end, seq1, seq2. The TRANSFORM effect strip is gone; every image or movie strip now has strip.transform and strip.crop. The timing properties frame_final_start, frame_final_end, frame_final_duration, frame_offset_start, frame_offset_end and frame_start are deprecated (removal planned for 6.0) in favour of left_handle, right_handle, duration, left_handle_offset, right_handle_offset and content_start.

Instructions

1. Initialize the sequence editor

import bpy

scene = bpy.context.scene
if not scene.sequence_editor:
    scene.sequence_editor_create()
strips = scene.sequence_editor.strips
scene.frame_start = 1
scene.frame_end = 250
scene.render.fps = 24

2. Add strips

# Video strip
strip = strips.new_movie(name="Clip_A", filepath="/srv/footage/interview_cam_a.mp4", channel=1, frame_start=1)
strip.left_handle_offset = 24    # trim start (skip 1 sec at 24 fps)
strip.right_handle_offset = 48   # trim end

# Image strip (hold for duration)
img = strips.new_image(name="Title_Card", filepath="/srv/footage/title_card.png", channel=2, frame_start=1, fit_method='FIT')
img.duration = 72  # 3 seconds at 24fps

# Image sequence
import glob
img_files = sorted(glob.glob("/srv/renders/shot_010/frame_*.png"))
img_seq = strips.new_image(name="Render", filepath=img_files[0], channel=1, frame_start=1)
for f in img_files[1:]:
    img_seq.elements.append(f.split("/")[-1])

# Audio strip with fade in/out
audio = strips.new_sound(name="Music", filepath="/srv/audio/theme_music.mp3", channel=3, frame_start=1)
audio.volume = 0.0
audio.keyframe_insert(data_path="volume", frame=1)
audio.volume = 0.6
audio.keyframe_insert(data_path="volume", frame=24)
audio.volume = 0.6
audio.keyframe_insert(data_path="volume", frame=audio.right_handle - 48)
audio.volume = 0.0
audio.keyframe_insert(data_path="volume", frame=audio.right_handle)

3. Transitions and effects

# Cross dissolve (strips must overlap and be on different channels)
cross = strips.new_effect(
    name="CrossDissolve", type='GAMMA_CROSS', channel=3,
    frame_start=clip_b.left_handle, length=24,
    input1=clip_a, input2=clip_b
)

# Color strip (solid background)
color = strips.new_effect(name="BlackBG", type='COLOR', channel=1, frame_start=1, length=48)
color.color = (0, 0, 0)

# Text overlay
text = strips.new_effect(name="Title", type='TEXT', channel=4, frame_start=1, length=72)
text.text = "Harbor Cities, Episode 3"
text.font_size = 80
text.color = (1, 1, 1, 1)
text.location = (0.5, 0.5)       # relative position in the frame
text.anchor_x = 'CENTER'         # text box anchor relative to location
text.anchor_y = 'CENTER'
text.alignment_x = 'CENTER'      # alignment of lines inside the box
text.use_shadow = True

# Speed control
speed = strips.new_effect(name="SlowMo", type='SPEED', channel=5,
    frame_start=clip_a.left_handle, length=clip_a.duration, input1=clip_a)
speed.speed_control = 'MULTIPLY'     # default STRETCH fits the input to the strip length
speed.speed_factor = 0.5

# Transform (position, scale, rotation) is a property of image/movie strips
clip_a.transform.scale_x = 1.2
clip_a.transform.scale_y = 1.2
clip_a.transform.offset_x = 40       # pixels
clip_a.crop.min_x = 20               # crop left edge by 20 px

4. Strip modifiers for color correction

strip = strips["Clip_A"]
bc = strip.modifiers.new(name="BrightContrast", type='BRIGHT_CONTRAST')
bc.bright = 0.1
bc.contrast = 0.15

cb = strip.modifiers.new(name="ColorBalance", type='COLOR_BALANCE')
cb.color_balance.lift = (0.95, 0.95, 1.0)
cb.color_balance.gain = (1.1, 1.05, 0.95)

5. Render the final video

scene = bpy.context.scene
scene.render.filepath = "/tmp/final_edit.mp4"
scene.render.image_settings.file_format = 'FFMPEG'
scene.render.ffmpeg.format = 'MPEG4'
scene.render.ffmpeg.codec = 'H264'
scene.render.ffmpeg.audio_codec = 'AAC'
scene.render.ffmpeg.audio_bitrate = 192
scene.render.resolution_x = 1920
scene.render.resolution_y = 1080

# Auto-set frame range from strips
all_strips = scene.sequence_editor.strips_all
if all_strips:
    scene.frame_start = min(s.left_handle for s in all_strips)
    scene.frame_end = max(s.right_handle for s in all_strips)

bpy.ops.render.render(animation=True)

Examples

Example 1: Assemble an edit from a shot list

User request: "Build a timeline from video clips with crossfades between them"

import bpy, os

clips = ["/srv/footage/shot_01.mp4", "/srv/footage/shot_02.mp4", "/srv/footage/shot_03.mp4", "/srv/footage/shot_04.mp4"]
crossfade_frames = 12

scene = bpy.context.scene
scene.render.fps = 24
if not scene.sequence_editor:
    scene.sequence_editor_create()
strips = scene.sequence_editor.strips
current_frame = 1
prev_strip = None

for i, clip_path in enumerate(clips):
    if prev_strip and crossfade_frames > 0:
        current_frame -= crossfade_frames
    strip = strips.new_movie(
        name=os.path.splitext(os.path.basename(clip_path))[0],
        filepath=clip_path, channel=1 + (i % 2), frame_start=current_frame
    )
    if prev_strip and crossfade_frames > 0:
        strips.new_effect(name=f"Fade_{i}", type='GAMMA_CROSS', channel=3,
            frame_start=current_frame, length=crossfade_frames,
            input1=prev_strip, input2=strip)
    current_frame += strip.duration
    prev_strip = strip

scene.frame_start = 1
scene.frame_end = current_frame

Example 2: Batch add text overlays from data

User request: "Add text titles at specific timecodes"

import bpy

titles = [("Introduction", 0, 3), ("Chapter 1", 15, 4), ("Chapter 2", 120, 4), ("Conclusion", 300, 5)]
fps = bpy.context.scene.render.fps
strips = bpy.context.scene.sequence_editor.strips

for text_content, start_sec, dur_sec in titles:
    start_frame = int(start_sec * fps) + 1
    text = strips.new_effect(name=text_content[:20], type='TEXT', channel=5,
        frame_start=start_frame, length=int(dur_sec * fps))
    text.text = text_content
    text.font_size = 60
    text.color = (1, 1, 1, 1)
    text.location = (0.5, 0.15)
    text.anchor_x = 'CENTER'
    text.alignment_x = 'CENTER'
    text.use_shadow = True

Guidelines

• Always call scene.sequence_editor_create() if scene.sequence_editor is None.
• Use strips.new_movie(), new_image(), new_sound(), new_effect(). Avoid bpy.ops.sequencer.* in background mode.
• Cross dissolves need strips on different channels with overlapping frames. Alternate channels (1, 2, 1, 2).
• left_handle_offset / right_handle_offset trim content without moving the strip; left_handle is where the strip starts on the timeline. The older frame_offset_*, frame_final_* and frame_start names still work in 5.x but are deprecated.
• For audio fades, keyframe strip.volume — no built-in fade effect in VSE.
• Render long edits as image sequences first, then assemble with ffmpeg.
• Set resolution to match source footage to avoid scaling artifacts, or pass fit_method (FIT, FILL, STRETCH, ORIGINAL) when adding image and movie strips.
• The container default of scene.render.ffmpeg.format is MKV; set MPEG4 explicitly for .mp4 output. ffmpeg.constant_rate_factor controls quality.
• VSE processes channels bottom-to-top — higher channels layer on top.